AIMIP Phase 1: systematic evaluations of AI weather and climate models
2026
Read AIMIP Phase 1: systematic evaluations of AI weather and climate models (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-06944.txt.
Description
Brian Henn and twenty co-authors launch AIMIP, a comparison project that tests AI weather and climate models under one common protocol. Borrowed from climate science tradition, the project fixes the experiment, output format, and training rules. Models train on historical reanalysis. They must simulate the atmosphere given prescribed historical sea surface temperatures from 1979 to 2024. Shared rules expose which framework and architecture choices actually change behavior.
Models are scored on biases, trends, response to El Niño patterns, temporal variability, and out-of-sample generalization. The AI models simulate past climate and forcing responses about as well as a conventional physics-based model. Some, however, underestimate past warming trends, and predictions diverge on the generalization tests. The phase one dataset is public for further evaluation.
Catalog record
| YEAR | 2026 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 72 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.06944